I was wondering just yesterday if a model of “why waste time say lot word when few word do trick” would be easier on the tokens. I’ll have to give this a try lol